• Corpus ID: 2151698

Limits of CSP Problems and Efficient Parameter Testing

@article{Karpinski2014LimitsOC,
  title={Limits of CSP Problems and Efficient Parameter Testing},
  author={Marek Karpinski and Roland Mark{\'o}},
  journal={ArXiv},
  year={2014},
  volume={abs/1406.3514}
}
We present a unified framework on the limits of constraint satisfaction problems (CSPs) and efficient parameter testing which depends only on array exchangeability and the method of cut decomposition without recourse to the weakly regular partitions. In particular, we formulate and prove a representation theorem for compact colored $r$-uniform directed hypergraph ($r$-graph) limits, and apply this to $r$CSP limits. We investigate the sample complexity of testable $r$-graph parameters, we… 
Explicit Bounds for Nondeterministically Testable Hypergraph Parameters
TLDR
This work provides the first effective upper bound on the sample complexity of any nondeterministically testable $r-uniform hypergraph parameter as a function of thesample complexity of its witness parameter for arbitrary $r$.
Complexity of Nondeterministic Graph Parameter Testing (Revised Version)
TLDR
The sample complexity of nondeterministically testable graph parameters is studied and existing bounds on it are improved by several orders of magnitude.
On the Complexity of Nondeterministically Testable Hypergraph Parameters
TLDR
The paper proves the equivalence of the notions of nondeterministic and deterministic parameter testing for uniform dense hypergraphs of arbitrary order and introduces a new notion of a cut norm for hyper graphs of higher order, and employs regularity techniques combined with the ultralimit method.
Complexity of Nondeterministic Graph Parameter Testing
TLDR
The sample complexity of nondeterministically testable graph parameters is studied and existing bounds on it are improved by several orders of magnitude.

References

SHOWING 1-10 OF 56 REFERENCES
Approximation of Global MAX-CSP Problems
TLDR
The existence of an efficient sampling method for the MAX-CSP class of problems with linear global constraints and bounded feasibility gap is proved and a polynomial in � 1 sample complexity bound is given for the first time.
Testing properties of graphs and functions
We define an analytic version of the graph property testing problem, which can be formulated as studying an unknown 2-variable symmetric function through sampling from its domain and studying the
Polynomial time approximation schemes for dense instances of NP-hard problems
We present a unified framework for designing polynomial time approximation schemes (PTASs) for “dense” instances of many NP-hard optimization problems, including maximum cut, graph bisection, graph
Random sampling and approximation of MAX-CSPs
Random sampling and approximation of MAX-CSP problems
TLDR
A new efficient sampling method for approximating r-dimensional Maximum Constraint Satisfaction Problems, MAX-rCSP, on n variables up to an additive error εnr, which gives for the first time a polynomial in ε—1 bound on the sample size necessary to carry out the above approximation.
Yet another algorithm for dense max cut: go greedy
TLDR
It is proved that when the vertices are considered in random order, the algorithms are still approximation schemes, and this work generalizes to dense maximum constraint satisfaction problems.
Approximation Schemes for the Betweenness Problem in Tournaments and Related Ranking Problems
We settle the approximability status of the Minimum Betweenness problem in tournaments by designing a polynomial time approximation scheme (PTAS). No constant factor approximation was previously
Approximation Algorithms for Graph Homomorphism Problems
TLDR
The maximum graph homomorphism (MGH) problem is introduced and a 0.8535-approximation algorithm for multiway uncut is presented based on a natural linear programming relaxation that implies an algorithm for distinguishing between certain average-case instances of the subgraph isomorphism problem that appear to be hard.
A new rounding procedure for the assignment problem with applications to dense graph arrangement problems
Abstract.We present a randomized procedure for rounding fractional perfect matchings to (integral) matchings. If the original fractional matching satisfies any linear inequality, then with high
Property testing and its connection to learning and approximation
TLDR
The authors study the question of determining whether an unknown function has a particular property or is /spl epsiv/-far from any function with that property, and devise algorithms to test whether a graph has properties such as being k-colorable or having a /spl rho/-clique.
...
...